Robust Fine-Tuning of Vision-Language Models for Domain Generalization
Published in 2023 IEEE High Performance Extreme Computing Conference (HPEC), 2023
This paper benchmarks the performance of vision-language foundation models on satellite hyperspectral imagery under limited data conditions and analyzes scaling laws for training foundation models using few-shot data samples. The work extends to detecting objects in X-ray imagery using natural language, enabling open-set detection.
Recommended citation: K. Vogt-Lowell*, N. Lee*, T. Tsiligkaridis, M. Vaillant. (2023). "Robust Fine-Tuning of Vision-Language Models for Domain Generalization." 2023 IEEE High Performance Extreme Computing Conference (HPEC). (* equal contribution)
Download Paper